Research On The Prediction Of The Remaining Life Of High-speed Train Cooling Fan Electromagnetic Contactor

The electromagnetic contactor of the cooling fan is one of the important switching elements of the high-speed train,and its performance gradually degrades with the increasing service time of the train.Once it fails to work,the cooling fan could not operate normally either,which would affect the performance on heat dissipation of the traction motor,even cause serious accidents.The electromagnetic contactor of the cooling fan has a long service life,and the cost on online monitoring the health status of the contactor is quite high.Although,regular replacement of contactor in real projects guarantees reliable operation of the contactor,the unreasonable replacement cycle causes much waste of the contactor’service life.In order to make full use of the contactor under prerequisite of train’s safe operation,it is important to research the residual life prediction of the contactor.To solve this problem,an electric-mechanical coupling dynamic simulation model of the contactor is established based on the failure mechanism.Meanwhile,the sensitivity of different characteristic parameters of the contactor on degradation of contact clearance is researched.An accelerated test program is designed and a test rig is built,which is used to test the characteristic parameters of the contactor in whole-life-cycle degradation process.A mapping model of degradation data from accelerated condition to conventional condition is established.Moreover,a nonlinear degradation model is established to predict the residual life of the contactor,and the estimating method of the model parameters is researched.Then,the nonlinear degradation model based on accelerated factor is established,and the updating method of the model parameters is researched.This thesis mainly contains following parts:1.The failure mechanism of the contactor and the effect from degradation of contact clearance on characteristic parameters are analyzed.Then,an electric-mechanical coupling dynamic simulation model of the contactor is established,which is used to simulate the pulling and releasing process of contactor dynamically under different contact clearances conditions.And this model is verified by a measured coil current curve.Moreover,the sensitivity of characteristic parameters on degradation magnitude of contact clearance is analyzed based on stimulation findings,and the results indicate that the super-path time has the most sensitivity to the contact clearance degradation.2.An acceleration test program for contactor is designed.The grades of accelerated stress and the test method are determined based on the accelerated test theory.Then,a test rig is built to monitor the contactor degradation process under different current conditions.Based on the acceleration model,a least square fitting analysis is applied on test data,and a mapping model of degradation data from accelerated condition to conventional condition is established.3.Different function models are used to make linear regression for the contactor degradation data,and an error analysis on prediction results of the regression model is carried out.After regression analysis,the linear,power and exponential functions degradation models are established respectively based on Wiener process,and a method combining maximum likelihood estimation and Genetic Algorithm-Simulated Annealing is proposed to estimate the model parameters.The probability density function of the contactor residual life distribution is calculated by the degradation model,which can be used to predict the residual life.In addition,the results of error analysis and comparison of prediction data of three kinds of degradation models manifest that residual life prediction method based on power-function nonlinear degradation model owns highest feasibility.4.To solve the problem of increasing prediction error from accelerated stress impacting the parameters of degradation model,the power function nonlinear degradation model based on accelerated factor is researched.To solve the problem of individual differences among samples,the updating method of model parameters is researched by using Particle Filter algorithm.The life prediction results show that the accuracy of residual life prediction is improved when the Particle Filter algorithm is used to update the model parameters.